Doped graphene and mxene heterostructures incorporated in polyvinylidene fluoride matrix for piezoelectric energy harvester and neuromorphic memristor applications

Abstract

The rapid advancement of artificial intelligence (AI) technologies has escalated the demand for sustainable, efficient energy harvesting and computing solutions. This study addresses critical challenges in energy consumption and computational power integration within AI systems by developing polymer-based piezoelectric nanogenerators (PENGs) and temperature-adaptive memristors using doped graphene and MXene-based 2D/2D heterostructures. PENGs were fabricated by incorporating nitrogen, sulfur, and phosphorus tri-doped graphene (NSPG) and Ti3C2Tx MXene heterostructures into a polyvinylidene fluoride (PVDF) matrix. While these devices effectively converted mechanical vibrations into electrical energy, their performance degraded at elevated temperatures. To overcome this, quasi-3D nano-heterogeneous configurations were optimized with a graphene-to-MXene ratio of 1:2, achieving a peak-to-peak (p-p) open-circuit voltage (Voe) of 9.1 V, an average short-circuit current (/se) of 1.51 mA, and a power density (Pd) of3.1 µW/cm2 at room temperature (RT). At 90 °C, the PENG exhibited Voe and lse of 24 V and 2.3 µA, respectively, with a Pd of 3.85 µW/cm2, demonstrating its suitability for high-temperature automotive applications. However, NSPG-ThC2Tx combinations lacked sufficient stability, limiting energy harvesting from low-intensity human motions at RT. Replacing ThC2Tx with ThCNTx MXene improved interfacial coupling, enhancing stability and output, with a p-p Voe of 14.6 V and Pd of 2.2 µW/cm2 at RT. Mechanistic analysis revealed strong interfacial interactions that enhanced PVDF's electroactive �-phase formation and synergistic mechanical resilience, boosting performance. Additionally, flexible, temperature-adaptive memristors were fabricated using NSPG-ThCNTx composites in Ag nanowire/nanocomposite/ITO configurations. These memristors mimicked biological synaptic functions, such as short- and long-term plasticity, spike-timing-dependent plasticity, and paired-pulse facilitation, with stable operation up to 50 °C. The stability was attributed to thermal synergy among NSPG, Ti3CNTx, and PVDF, supported by Ag ion migration and redox reactions forming conductive filaments. These advancements highlight the potential of these devices for neuromorphic computing and artificial neuron regulation in complex environments.